An American Remarriage Table - Casualty Actuarial Society

71 Pages·2001·2.77 MB·English

279
AN AMERICAN REMARRIAGE TABLE
AN AMERICAN REMARRIAGE TABLE
BY
~VILLIAI~ F. ROEBER AND RALPH M. MARSHALL
In many lines of casualty insurance the entire amount of the
loss is due and payable immediately upon its determination. In
workmen's compensation insurance, however, it is customary for
the losses to be paid in installments over an extended period, and
in certain instances the payment is contingent upon the benefi-
ciary's remaining alive and in the same conjugal condition. This
is especially true of benefits in fatal cases. The compensation
laws of the majority of the states provide that in the case of a
fatal accident arising out of and in the course of employment, the
widow of the deceased employee shall be entitled to compensation,
payable periodically over a number of years, usually varying from
four to eight, and the payment of such compensation is contingent
upon the widow remaining alive and unmarried during the period.
In New York and certain other states, such payments continue
during the unmarried life of the widow. In order to set up proper
reserves, compute present values in case of commutation of
future payments to a lump sum, and carry out other calculations
in regard to incurred losses, it is necessary to have information
regarding death and remarriage probabilities.
There are a number of mortality tables in use at the present
time, the most important in the workmen's compensation insur-
ance field being the American Experience Table and the United
States Life Tables. The former is based upon the experience of a
large life insurance company and was first published in 1868. In
some states the use of this table in determining any lump sum
settlements payable under the workmen's compensation law is
required by statute. The United States Life Tables are based
upon census returns and contain separate tabulations regarding
male and female lives. The latest complete set are based upon
the census of 1910. These tables, being derived from general
population statistics, are perhaps more representative of condi-
tions found among beneficiaries under the various workmen's
compensation acts than any table based upon life insurance
280 NA NACIREMA EGAIRRAMER ELBAT
statistics, the mortality rate of which would be influenced by
selection of desirable risks through medical examination.
However, when we come to the question of the rate of remar-
riage, there has been, up to the present time, no information from
American sources available upon this subject; due, no doubt, to
the fact that until comparatively recent years, there was no
necessity for such information. When the adoption of workmen's
compensation legislation made such statistics desirable, it was
necessary to resort to foreign sources. The Insurance Depart-
ments of New York and Missouri have issued tables for commut-
ing compensation benefits to a present value basis in fatal cases.
The remarriage contingency of these tables depends upon statis-
tics of the Dutch State Insurance Fund which administers the
Workmen's Compensation Laws of Holland, the fatal benefits of
which are much the same as those provided by the Compensation
Act of the State of New York. These remarriage data have been
in general use in the United States in connection with compen-
sation cases.
Although a comparison of the table based upon the Dutch data
with a limited volume of early American remarriage experience
showed approximately the same average results,* the continued
use of this table is open to question. During recent years there
has been a growing feeling that we should have a remarriage
table based upon United States statistics. The Dutch Table is
based upon European statistics; more extensive American data
may yield different results. The period covered was the latter
part of the 19th century; customs may have changed consider-
ably since then. The remarriage rates given are ultimate rates
depending upon the widow's age ; during the early years of widow-
hood, the duration of widowhood may have a greater effect upon
the remarriage rate than the widow's age. No account is taken
of the number of dependent children; possibly the number of
children will have an influence on the remarriage rate. The con-
struction of an American Remarriage Table was considered at
various times by the United States Department of Labor, by the
New York Department of Labor, and also by the Casualty Actu-
arial Society. In 1929 the development of such a table was
* See---"Remarriage Experience of Pennsylvania Compensation Insurance
Carriers Policy Years 1916-1919", by E. H. Downey, Proceedings, Vol.
VIII, Page 201; and Written Discussion by Mr. M. M. Dawson, Proceed-
b2gs, Vol. IX, Page .701
AN A~ERICAN RE,CARRIAGE TABLE 281
undertaken by the Society and the project was placed in charge
of a committee consisting of seven members.
In considering sources from which the basic data could be
obtained, the Committee recognized two possibilities. The data
might be obtained either from industrial commissions or other
state departments charged with the administration of the com-
pensation laws, or directly from the insurance carriers. Further
investigation showed that available state department records were
somewhat limited in geographical distribution and in most in-
stances were kept in such manner that the information required
was not readily available. Accordingly, it was decided to secure
the information directly from the insurance carriers' records,
supplementing these data with experience obtained where pos-
sible from state departments, rating bureaus and monopolistic
state funds.
A rough survey was made of the volume of experience which
had developed since the adoption of compensation acts in the
United States and it was estimated that between 17,000 and
18,000 cases involving widows would be available.
Inasmuch as most insurance carriers are accustomed to report
their workmen's compensation experience to the National Council
on Compensation Insurance, it was considered desirable to have
the Couhcil serve as the collecting agency for the remarriage
data. The National Council agreed to gather the data and to
perform all the clerical work necessary in the construction of a
remarriage table. On November 15, 1929, the Call for Remar-
riage Data was issued. This call was sent to all members of the
National Council and was supplemented by letters to state com-
pensation insurance funds in monopolistic states and to carriers
not affiliated with the National Council, requesting their cooper-
ation and assistance. A supply of blank forms for reporting the
experience was furnished to each cooperating organization. A
copy of the form with detailed instructions for preparation of the
report is appended as Exhibit I.
Before issuing the Call for Remarriage Data, the compensation
laws of the various states were examined for conditions whi.ch
might affect the accuracy of the carriers' claim records regarding
remarriage. In some states the rate of compensation is not
affected by remarriage and in other states the compensation is
unaffected by remarriage when there are dependent children. In
282
AN AMERICAN REMARRIAGE TABLE
such jurisdictions the insurance carriers have no vital reason to
accurately record changes in the marital status of the widow.
Accordingly the call was limited to fatal cases coming under
compensation laws which provide for a material change in the
benefits upon the widow's remarriage. The data were requested
for the following jurisdictions: with limitations as indicated:--
REPORTS LIMITED TO CASES
Resulting From
Jurisdiction Accidents Involving
Occurring On
And After
Alabama
Arizona November ,3 5291
Colorado Widow with no other dependents
Connecticut Widow with no other dependents
District of Columbia
Georgia
Hawaii
Idaho
Illinois July ,1 7291 Widow with no other dependents
Indiana
Iowa July I, 7191 Widow with no other dependents
Kansas March ,21 3191
Kentucky
Louisiana
Maine July ,8 1291 Widow with no rea~to dependents
Maryland Widow with no other dependents
Massachusetts tsu-guA ,01 2291
Michigan September ,5 7291 Widow with no other dependents
Minnesota Widow with no other dependents
Missouri Widow with no other dependents
Montana
New Jersey April ,1 3191
New Mexico
New York
Rhode Island April l, 6291
South Dakota Widow with no other dependents
Tennessee Widow with no other dependents
Utah Widow with no other dependents
U. .S Longshore-
men's Act
Vermont
Virginia Widow with no other dependents
Except as noted above a report was requested on each fatal
case, involving a dependent widow, arising at any time since the
inception date of the compensation act. In states where informa-
tion was desired only on cases arising after specified dates, the
compensation laws prior to those dates were of such nature that
AN AMERICAN REMARRIAGE TABLE 283
benefits to the widow were not materially altered by her
remarriage.
The reporting agencies were asked to advise the National
Council of any period for which their files did not contain the
records for all reportable cases arising during that period. The
carriers were also requested to report every case, showing all
available information, even though their files were incomplete
with regard to one or more items called for.
It will be noted from the blank form provided for reporting
these statistics that the following items are listed:
Reporting Carrier
.1 Husband's Name
.2 Widow's Name
.3 Identification No.
4. Policy Year
5. State
.6 Classification
.7 Date of Husband's Death (Month, Day, Year)
.8 Date of Widow's Birth (or age at husband's death)
.9 Date of Widow's Death
.01 Date of Widow's Remarriage
.11 (I) Date of Termination of Widow's Benefit other than (9)
or (10)
(II) Mode of Termination
.21 Date Status of Case was Last Observed---Open Cases
.31 Number of Dependent Children at Date of Husband's Death.
The name of the Carrier and the first six items were requested
primarily for identification purposes. Item 4 "Policy Year"
allowed of a division of the cases by period of occurrence. Item 5
--"State" was included to permit an analysis by geographical
division, and Item 6--"Classification" to provide the basic infor-
mation for a study by kind of industry.
The remaining items are the essential ones which were used in
the calculation of the remarriage rates. In order to give a clearer
idea of the nature and purpose of these items it is desirable to
present a definition of a remarriage rate and to outline the general
principles of calculation.
For the purposes of this investigation, the remarriage rate is
284 AN AMERICAN RE't,A'ARRIAGE TABLE
defined as the probability that any specific widow, considered at
a definite date, will remarry within one year from that date. In
determining the remarriage rate, the Committee decided that the
influence of the widow's age, the period of widowhood, the number
of dependent children, the geographical division, and the type of
industry in which her husband was engaged at time of the fatal
injury, would be studied.
In general the method employed in determining the remarriage
rate was to study the remarriage history of a large group of
widows of the same age under observation for a period of one
year. The remarriage rate was taken as the ratio of the number
of remarriages to the total number under observation. Actually
each case was observed for as long a period as possible and adjust-
ments were made for duration of widowhood, and for withdrawals
from observation due to death, end of legal period, or other causes.
As an initial step in the calculation of remarriage rates it was
necessary to ascertain from the reports the following facts :--
(1) the widow's age at husband's death,
(2) the cause of withdrawal from observation, and
(3) the period of observation.
The widow's age at husband's death was determined in many
cases directly from Item ,8 which provides that in the event the
exact date of the widow's birth is unknown, her age at husband's
death shall be reported. Where the date of widow's birth was
given, it was necessary to calculate the widow's age by computing
the time from the date of her birth to the date of husband's
death. It is customary for a person in stating his age to give it
as of his last birthday. In order to keep the data homogenous,
the widow's age at her birthday immediately preceding the date
of her husband's death was calculated in cases where the date
of birth was given. It was realized that this procedure would
result in a remarriage table in which the tabulated ages would be
on the average one-half year under the actual ages. However, no
error will result if in the use of the table the widow's age at last
birthday is used to enter.
The date and cause of withdrawal from observation were
determined from the information reported under Items 9 to 21
inclusive. If the status of the widow remained unchanged dur-
ing the period of observation, this fact was indicated by an entry
AN AMERICAN REMARRIAGE TABLE 285
for Item 12 with Items 9, 10 and 11 blank. The following code
was adopted to indicate the cause of withdrawal:-
Case still open at end of period ............................ 0
Death of widow ................................................ 2
Remarriage of widow ............................................... 2
Lump sum settlement ......................................... 3
End of legal period .......... - ................................. 4
Claim disallowed ................................................ 5
Any other ............................................................... 6
The period of observation was determined by calculating the
elapsed time between the date of husband's death and the date
of the last observation. On closed cases the date of last observa-
tion would be the date of the widow's death or" remarriage, or
the date of the final compensation payment. On open cases the
date of last observation was that entered under Item 12. The
period of observation was calculated in years and nearest whole
months.
As the reports were received at the National Council they were
audited and the required calculations made. The widow's age,
the cause of withdrawal and the period of observation were noted
on each report.
In cases where there were conflicting notations or apparent
errors in the information given, the carriers were communicated
with to ascertain the required corrections; and, in cases where
some of the essential data were lacking, an attempt was made to
secure the missing information.
The basic data were obtained from the individual case reports
of the insurance carriers. The Pennsylvania data for all carriers
were prepared by the Pennsylvania Compensation Rating and
Inspection Bureau and furnished to the National Council. The
New Jersey Compensation Rating and Inspection Bureau assisted
in securing the data for New Jersey from carriers which are not
members of the National Council, and the North Dakota Work-
men's Compensation Bureau reported the data for that state.
For all other states the insurance carriers reported directly to the
Council. Considerable work was required in obtaining the neces-
sary information from the files, and not all carriers were able to
give the matter their immediate attention. As a result there was
some delay in filing these reports and the returns were not all
received until early in 1931.
286
AN AMERICAN REMARRIAGE TABLE
In the meantime consideration was given to methods of tabu-
lating and compiling the data submitted. It was felt that the
use of punch cards for recording these data would be of great
advantage, as the mechanical sorting would permit the data to be
quickly assembled in any order desired, and the printer tabu-
lator would allow the presentation of figures in the order decided
upon and would facilitate the obtaining of sub-totals. Accord-
ingly, a punch card for recording the remarriage data was drawn
up and approved by the Committee on Remarriage Table. This
punch card was arranged to record all of the essential informa-
tion given on the remarriage form report. A facsimile of the
card is given on page 287.
It will be'noted that the items recorded are :--
1. Carrier Code Number
2. Policy Year
3. State Code Number
4. Classification and Schedule Code Number
5. Date of Husband's Death
6. Widow's Age at Husband's Death
7. Period of Observation--Years and Months
8. Mode of Withdrawal Code Number
9. Number of Dependent Children
10. Counter (Always Punched "1")
11. Case Serial Number
Standard codes for "Carrier" and "State" already existed and
were adapted to this recording by the necessary additions. The
numerical code, which was used to indicate the reason for with-
drawal from observation, has already been given. The remaining
items were numerical items and could be transferred to the punch
card directly. The "counter" column was included as an aid in
the summary of tabulations and was punched "1" for each case.
The last item, "Case Serial Number," was included as an aid in
identification in case it should be found desirable to refer back to
the original report from the punch card. This serial number was
assigned to the original report at the National Council.
It will be noted that the items recorded on the punch cards
permit studies fO remarriage to be made by state, by industry
schedule, and by period.
The reports filed covered the experience for policy years 1911
to 1929 inclusive. A preliminary review of the data revealed that
IYCILOP LAUNAM SSALC ETAD FO wmowsl PERIOD'OF ~J <=., .ON FO
CARRIER YEAR S'rATE CODE ! I SCHED. S'DNABSUH .o I o,Y HTAED ~R ,E~,",;"s'1 ~ ,i I"O'TAVRES80I SRY I"OS. '~ .DL I.HPCEO NUMBSEIE RIAL l >
3 0 010 OtO 010 0 0 0;0 0 .o.oi? oLO O " o olo olo ol 000 ;R22RR 2RR22 R2
....................... i .... I
1 1 I11 I11 I11 1 1 1 1 1 1 111 111 1 1 111 111 1 111 1
2221221221222222 212 212 2 2 2;'2 212 2: 222 2 2122222222222222 2
I
i 313 313 3 3 313 "313 3i 333 3 3;333333333333333 j..p
...... .... i .... i ........ ..................... i-..~ .. ~Z
¢ 4 414 414 414 4 4 4 4 414 4 4 414 4i4 4: 44-4 4 4,444
!
!
5551551551555555 5 ,515 5 5 515 515 5 555 5 51555555555555555
6 6 616 616 616 6 6 6 6 6 6 1616 6 6 616 616 6 666 6 6,666! 666666666666
y- -y
Y ..7:7' ...... 7 7 Y Y YY YT Y
7 7 717 717 717 7 7 7 7 7 7 717 7 7717 717 7;
! J'P
888188188888888 8 818 8 8 818"818 8 888 8 81888888888888888
!
9 9 919 919 919 9 9 9 9 9 9 919 919 919 9i9 19 999 9:9 9 9 ~ ~ 9 9 ~ 9 ~
I 2 3 I 4 5 I ~ .7 I 8 9 10 tl 1~ 13 411 1~)116 1711~ 19 :~0 21124? ;~|24 2~il 27 20 30 31 32 ~ ~ 40 41
O,'L
0O
q-.
288
AN ANIERICAN REMARRIAGE TABLE
in many companies, claim files on closed cases had been destroyed
for the early policy years and the required data could only be
reported on cases that were still open. Inclusion of such cases
as were reported for the earlier period would give a false basis of
exposure and tend to distort the results. Therefore, it was de-
cided that the study would be confined to data reported for
policy years 1921 to 1929 inclusive.
Exhibit II presents a summary of the volume of the data which
serves as the basis for the calculated remarriage rates. The
number of cases observed, numbe~ of remarriages, and total
exposure are given by age groups. The completed tables are based
on 10,699 cases representing a total exposure of 37,040 years.
A further review of the data indicated that the remarriage rate
varied more with the year of widowhood during the first few
years than with the age of the widow. It was decided to prepare
a select table showing remarriage rates by age of widow for each
of the first six years of widowhood, the values for the sixth year
to be considered as ultimate.
It was evident that the data were too limited to give depend-
able remarriage rates when each year of widowhood was sepa-
rated into individual ages. In order to obtain a more substantial
volume of data, it was decided to calculate remarriage rates for
each individual age by using, under each year of widowhood, a
five year moving average centered at the mid-age of the five year
age period. For example the average remarriage rates for ages
61 to 20 inclusive would be taken for age 18; the average rates
for 12 to 52 inclusive would be taken for age 23, etc. Then by
considering a different grouping of ages, for example 71 to ,12 22
to 26, etc., the average rates for ages ,91 24, 29, etc., would be
obtained. By continuing this process under each year of widow-
hood for five different tabulations a moving average of the
remarriage rates for each consecutive age would be obtained.
In order to carry out this project, the punch cards were sorted
according to widow's age at husband's death and were assembled
in five year age groups. The cards for each age group were then
sorted by "period of observation," (i.e. "duration of widowhood").
Finally, the "withdrawals" from observation during each year of
widowhood were arranged according to "cause of withdrawal."
The cards so arranged were then run through the printer tabu-
lator and sub-totals obtained for each age group showing the total